Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
00:00:06,600 --> 00:00:09,366
hello and welcome back to nuke 2 3 1
2
00:00:09,366 --> 00:00:11,866
nuke compositing best practices
3
00:00:11,866 --> 00:00:13,133
this is Victor Perez
4
00:00:13,133 --> 00:00:14,099
and in this class
5
00:00:14,100 --> 00:00:15,600
we are going to discuss
6
00:00:15,600 --> 00:00:18,833
how to blend pre multiplyed images
7
00:00:18,866 --> 00:00:20,599
without destroying
8
00:00:20,600 --> 00:00:23,666
the grainy structure in the soft areas
9
00:00:23,666 --> 00:00:28,866
okay so let's start by presenting the most common
10
00:00:28,966 --> 00:00:30,066
case scenario
11
00:00:30,066 --> 00:00:32,899
which is a premortified CG
12
00:00:33,366 --> 00:00:36,733
to be mixed with a live action image
13
00:00:36,733 --> 00:00:38,299
so as you know
14
00:00:38,300 --> 00:00:39,400
what you are going to do
15
00:00:39,400 --> 00:00:43,333
is to match the grain from the original plate
16
00:00:43,333 --> 00:00:46,399
and apply it to the CG image
17
00:00:46,500 --> 00:00:47,700
okay in this class
18
00:00:47,700 --> 00:00:51,100
we are not going to discuss how to match the grain
19
00:00:51,166 --> 00:00:54,599
what I'm interested here is to understand
20
00:00:54,700 --> 00:00:58,266
how the grain is going to blend in the soft areas
21
00:00:58,266 --> 00:01:03,199
or even in other areas that are not in the alpha
22
00:01:03,266 --> 00:01:08,099
okay so let's start by understanding what is actually
23
00:01:08,300 --> 00:01:10,966
an application of the grain well
24
00:01:10,966 --> 00:01:13,699
let me tell you that applying the grain to an image
25
00:01:13,700 --> 00:01:16,966
is exactly the same as applying a color correction
26
00:01:16,966 --> 00:01:19,999
so this is my CG image
27
00:01:20,000 --> 00:01:24,166
okay so what I'm doing in here is applying grain
28
00:01:24,300 --> 00:01:27,600
and what this grain is actually doing
29
00:01:27,600 --> 00:01:31,866
is randomizing the frequency of those values
30
00:01:31,933 --> 00:01:35,633
so let's understand the concept of frequency
31
00:01:36,166 --> 00:01:37,566
to continue further
32
00:01:37,566 --> 00:01:40,166
so imagine that in here
33
00:01:40,466 --> 00:01:43,333
we have just a gradient okay
34
00:01:43,333 --> 00:01:45,733
it's just a linear gradient
35
00:01:45,733 --> 00:01:48,633
from zero value to value one
36
00:01:48,800 --> 00:01:49,466
in this case
37
00:01:49,466 --> 00:01:52,999
we are not displaying the gradient linearly
38
00:01:53,000 --> 00:01:56,166
because we are applying an s R
39
00:01:56,166 --> 00:01:59,699
G B call or look up to the beware process
40
00:01:59,700 --> 00:02:02,700
okay so if I deactivate that input
41
00:02:02,700 --> 00:02:04,566
known as you can see now
42
00:02:04,566 --> 00:02:09,299
is clearly a gradient from 0 to 1
43
00:02:09,300 --> 00:02:14,133
okay so let's analyze that using a slice 2
44
00:02:14,133 --> 00:02:18,399
okay you can find this is slice 2 on Wikipedia
45
00:02:18,900 --> 00:02:21,733
this has been created by Frank Rooter
46
00:02:21,733 --> 00:02:25,833
so what this is doing is actually analyzing every pixel
47
00:02:25,966 --> 00:02:28,566
below this yellow line
48
00:02:28,966 --> 00:02:30,766
and is putting the
49
00:02:30,966 --> 00:02:34,366
x coordinates as the special coordinates
50
00:02:34,366 --> 00:02:38,399
so left in here is left to the x
51
00:02:38,600 --> 00:02:45,100
and right here is the right of the x axis okay
52
00:02:45,100 --> 00:02:50,000
while the y axis is representing 0
53
00:02:50,000 --> 00:02:52,000
and the top is representing 1
54
00:02:52,000 --> 00:02:54,766
so this side is normalized
55
00:02:54,766 --> 00:02:57,466
okay so the higher is the value
56
00:02:57,466 --> 00:02:59,666
the higher in position of
57
00:02:59,666 --> 00:03:03,899
why is going to be the pixel represented in this line
58
00:03:03,900 --> 00:03:06,333
okay as you can see clearly
59
00:03:06,333 --> 00:03:09,066
this is a linear ramp okay
60
00:03:09,800 --> 00:03:14,033
so what if I apply the grain okay
61
00:03:14,133 --> 00:03:15,033
let's have a look
62
00:03:15,466 --> 00:03:16,999
well as you can see
63
00:03:17,066 --> 00:03:20,666
the overall ramp is still there
64
00:03:21,100 --> 00:03:23,266
what is actually happening is
65
00:03:23,266 --> 00:03:25,666
every pixel below the line
66
00:03:25,666 --> 00:03:30,299
is now getting a randomization because of the grain
67
00:03:30,300 --> 00:03:32,000
okay so in here
68
00:03:32,000 --> 00:03:33,200
as you can see
69
00:03:33,700 --> 00:03:38,866
you are finding tiny directions up and down
70
00:03:39,333 --> 00:03:42,566
on every single pixel of the RAM
71
00:03:42,566 --> 00:03:45,066
so if I go to the grain amount
72
00:03:45,466 --> 00:03:47,633
and I increase the amount
73
00:03:47,700 --> 00:03:50,633
what I'm doing is adding more contrast
74
00:03:50,666 --> 00:03:54,733
from the highest to the lowest part of the grain
75
00:03:54,733 --> 00:03:58,466
okay so I'm going to reset that to 1
76
00:03:58,500 --> 00:04:02,966
and that is the amplitude of the frequency
77
00:04:03,000 --> 00:04:04,166
but actually
78
00:04:04,166 --> 00:04:06,066
the frequency is
79
00:04:06,666 --> 00:04:09,066
the change that you are getting
80
00:04:09,066 --> 00:04:12,533
from one point of the curve to another
81
00:04:12,533 --> 00:04:16,433
so in here the frequency is quite high
82
00:04:16,700 --> 00:04:19,566
but if I apply a blur
83
00:04:19,566 --> 00:04:20,599
for instance
84
00:04:21,400 --> 00:04:23,466
see that now
85
00:04:23,866 --> 00:04:24,899
you can see
86
00:04:24,900 --> 00:04:27,833
this is like a wave now okay
87
00:04:28,366 --> 00:04:30,633
if I remove the blur
88
00:04:30,700 --> 00:04:31,733
you are going to see like
89
00:04:31,733 --> 00:04:34,133
the values are going up and down
90
00:04:34,133 --> 00:04:34,666
up and down
91
00:04:34,666 --> 00:04:35,333
up and down
92
00:04:35,333 --> 00:04:38,199
without the shape of the wave
93
00:04:38,200 --> 00:04:39,266
as before okay
94
00:04:39,266 --> 00:04:40,299
even sometimes
95
00:04:40,300 --> 00:04:41,433
you can finding here
96
00:04:41,500 --> 00:04:42,566
but have a look at this
97
00:04:42,566 --> 00:04:44,633
happening to the actual image
98
00:04:44,733 --> 00:04:48,966
see the more I increasing
99
00:04:50,566 --> 00:04:51,366
the blur
100
00:04:52,766 --> 00:04:54,333
the value in here
101
00:04:54,333 --> 00:04:58,466
of that randomization is being lost okay
102
00:04:58,533 --> 00:05:02,866
because the blur is actually a high frequency filter
103
00:05:02,866 --> 00:05:06,033
okay but have a look at the ramp
104
00:05:06,200 --> 00:05:08,666
before and after the process
105
00:05:08,700 --> 00:05:11,566
see the RAM is actually the same
106
00:05:11,966 --> 00:05:14,233
well quite close okay
107
00:05:14,600 --> 00:05:16,266
but it's not changing
108
00:05:16,266 --> 00:05:19,699
that is because the grain is not actually changing
109
00:05:19,700 --> 00:05:22,700
the predominant colour of the image
110
00:05:22,700 --> 00:05:24,533
and that is supposed to be like that
111
00:05:24,533 --> 00:05:27,333
if you are changing the colour of an image
112
00:05:27,333 --> 00:05:28,533
because of the grain
113
00:05:28,533 --> 00:05:30,466
then you are in trouble okay
114
00:05:30,466 --> 00:05:33,899
so the grain is actually creating a
115
00:05:33,966 --> 00:05:35,566
Saildo randomization
116
00:05:35,566 --> 00:05:37,966
what I mean with Saildo randomization
117
00:05:37,966 --> 00:05:39,133
is because in computer
118
00:05:39,133 --> 00:05:42,099
you cannot create fully randomize
119
00:05:42,100 --> 00:05:43,733
like the chaos okay
120
00:05:43,733 --> 00:05:45,799
you are creating a pattern
121
00:05:46,000 --> 00:05:48,900
that is looking like a randomization
122
00:05:48,900 --> 00:05:50,166
and this is very important
123
00:05:50,166 --> 00:05:52,833
because sometimes we are going to need to get
124
00:05:52,933 --> 00:05:55,399
the exact randomization
125
00:05:55,400 --> 00:05:56,666
from this note
126
00:05:56,700 --> 00:05:58,166
and from another note
127
00:05:58,166 --> 00:06:01,099
that we are going to see later in this class
128
00:06:01,100 --> 00:06:04,266
okay so I'm going to delete this blur
129
00:06:04,866 --> 00:06:09,733
so you know now that this is just a color correction
130
00:06:09,733 --> 00:06:15,533
so one of the main rules of compositing is
131
00:06:15,533 --> 00:06:18,999
every time you are applying a color correction
132
00:06:19,200 --> 00:06:21,700
to promote applied image
133
00:06:21,966 --> 00:06:27,499
you always have to apply the color correction on
134
00:06:27,800 --> 00:06:30,600
in an unpremorted state okay
135
00:06:30,600 --> 00:06:35,833
so we am pre multiply the image as return with the srgb
136
00:06:36,333 --> 00:06:38,599
then we apply the grain
137
00:06:38,866 --> 00:06:45,099
okay and then we pre multiply the image again okay
138
00:06:45,100 --> 00:06:46,800
so that way
139
00:06:46,800 --> 00:06:51,566
we are preserving every constant uh
140
00:06:51,566 --> 00:06:53,699
grain in every pixel
141
00:06:53,700 --> 00:06:57,166
even when we are getting semi transparent areas
142
00:06:57,166 --> 00:07:02,766
okay well sometimes you can find that notes
143
00:07:02,766 --> 00:07:04,866
like the grain in here
144
00:07:04,866 --> 00:07:07,166
grain have this
145
00:07:07,166 --> 00:07:09,266
apply only through alpha
146
00:07:09,933 --> 00:07:14,833
what this mean is quite similar to the ampromode
147
00:07:15,066 --> 00:07:17,666
ampromode then after the grain
148
00:07:17,733 --> 00:07:19,733
but it's not exactly the same as
149
00:07:19,733 --> 00:07:23,866
it's not the same to use a mask for a color correction
150
00:07:23,866 --> 00:07:27,066
or to use the alpha in the unpremot obligation
151
00:07:27,066 --> 00:07:29,299
and then premot obligation okay
152
00:07:29,300 --> 00:07:30,500
but for the green
153
00:07:30,500 --> 00:07:33,700
this rule has 13 exceptions
154
00:07:33,700 --> 00:07:36,400
because when you are actually getting
155
00:07:37,100 --> 00:07:39,266
the semi transparent areas
156
00:07:39,266 --> 00:07:40,766
what you are doing
157
00:07:40,800 --> 00:07:44,866
is mixing the grain from the original plate
158
00:07:45,133 --> 00:07:46,766
okay the grain in there
159
00:07:46,933 --> 00:07:50,166
with the grain on your CG
160
00:07:50,666 --> 00:07:52,266
and you know
161
00:07:52,866 --> 00:07:54,599
green plus grain
162
00:07:54,600 --> 00:07:57,700
is not equals twice the grain
163
00:07:58,300 --> 00:08:00,600
but it's actually the other way around
164
00:08:00,600 --> 00:08:02,233
when you are mixing grain
165
00:08:02,966 --> 00:08:05,366
you're getting actually less grain
166
00:08:05,533 --> 00:08:07,366
so let me show you here
167
00:08:07,366 --> 00:08:09,299
for instance this rum
168
00:08:09,300 --> 00:08:11,766
which is exactly the same as before I use
169
00:08:11,766 --> 00:08:12,766
yes another note
170
00:08:12,766 --> 00:08:14,399
but this is
171
00:08:15,500 --> 00:08:16,500
this lies too
172
00:08:16,500 --> 00:08:17,766
as before now
173
00:08:17,766 --> 00:08:19,366
is way clear
174
00:08:19,366 --> 00:08:21,299
that you are having randomization
175
00:08:21,366 --> 00:08:22,399
as you can see
176
00:08:23,133 --> 00:08:24,466
okay and here
177
00:08:25,266 --> 00:08:26,266
is the same
178
00:08:26,866 --> 00:08:30,366
but have a look what's happening when I'm
179
00:08:31,100 --> 00:08:33,833
averaging this and this grain
180
00:08:34,400 --> 00:08:37,966
and notice that they both have a different seed
181
00:08:37,966 --> 00:08:41,766
so they are actually the same pattern of the grain
182
00:08:41,766 --> 00:08:43,366
but in different moments
183
00:08:43,366 --> 00:08:47,699
okay so they are not at the same randomization time
184
00:08:47,700 --> 00:08:50,700
okay so if I average those
185
00:08:51,066 --> 00:08:52,433
and I have a look at that
186
00:08:52,933 --> 00:08:54,633
have a look at this curve
187
00:08:57,400 --> 00:08:58,200
see
188
00:08:59,500 --> 00:09:01,566
it's clear in here
189
00:09:02,000 --> 00:09:04,900
when I deactivate the average
190
00:09:05,733 --> 00:09:07,399
the grain expands
191
00:09:07,600 --> 00:09:10,400
but you don't need to have this
192
00:09:10,400 --> 00:09:11,900
is licensed to notice it
193
00:09:11,900 --> 00:09:13,466
you can just go here
194
00:09:14,500 --> 00:09:18,300
and notice that when I am activating the average
195
00:09:19,300 --> 00:09:22,766
the grain seems to slightly disappear
196
00:09:23,133 --> 00:09:25,533
um this is actually one of the tricks
197
00:09:25,533 --> 00:09:29,366
to create a plate without the grain
198
00:09:29,366 --> 00:09:32,999
you just blends the plate if the plate is not moving
199
00:09:33,000 --> 00:09:35,600
so you are going to get everything
200
00:09:35,600 --> 00:09:37,100
with all the frequencies
201
00:09:37,100 --> 00:09:39,766
of the grain just being average
202
00:09:39,766 --> 00:09:43,399
so they are going to disappear at the end of the day
203
00:09:43,400 --> 00:09:48,066
okay so I put another example in here with this
204
00:09:48,266 --> 00:09:51,966
using four grains and see
205
00:09:53,366 --> 00:09:56,833
the more you are going to to add to the average
206
00:09:56,900 --> 00:10:00,466
the more you are going to restore the original shape
207
00:10:00,666 --> 00:10:01,599
of the rump
208
00:10:01,666 --> 00:10:06,199
okay so the frequencies are fighting against each other
209
00:10:06,200 --> 00:10:11,033
okay so this is the problem of mixing the grain
210
00:10:11,066 --> 00:10:14,166
so that's why I'm going to spend this glass
211
00:10:14,166 --> 00:10:16,966
just to show you how is the right way
212
00:10:16,966 --> 00:10:20,866
or the right ways of creating the grain
213
00:10:20,866 --> 00:10:24,299
when you are applying mainly soft patches
214
00:10:24,300 --> 00:10:27,266
okay so in here
215
00:10:28,500 --> 00:10:33,600
if I apply the alpha instead of the primotification
216
00:10:33,600 --> 00:10:35,900
the result is going to be quite close
217
00:10:36,000 --> 00:10:38,333
still not the same well
218
00:10:38,333 --> 00:10:42,633
sometimes is gonna work this way with the amprimote
219
00:10:42,900 --> 00:10:44,200
and then the promote
220
00:10:44,200 --> 00:10:48,000
so the classic color correction amprimote application
221
00:10:48,133 --> 00:10:50,666
and then sometimes you are going to find
222
00:10:50,666 --> 00:10:53,833
that is going to be better just to apply a mask
223
00:10:53,933 --> 00:10:54,699
in this case
224
00:10:54,700 --> 00:10:57,266
if I want to create the same effect
225
00:10:57,266 --> 00:11:00,366
of the primo application set with the Alpha
226
00:11:00,366 --> 00:11:02,166
I'm going to use a shuffle
227
00:11:03,766 --> 00:11:07,266
in here I'm going to call the postage stamp
228
00:11:07,466 --> 00:11:08,666
and now I'm going to put
229
00:11:09,000 --> 00:11:11,900
for the sake of visualizing everything
230
00:11:12,700 --> 00:11:14,500
this in the alpha
231
00:11:14,500 --> 00:11:16,700
okay so this is the alpha
232
00:11:17,333 --> 00:11:20,299
and now I'm going to deactivate that
233
00:11:20,300 --> 00:11:23,200
because I don't want to apply twice the same effect
234
00:11:23,466 --> 00:11:24,966
and now I'm going to call
235
00:11:26,166 --> 00:11:27,299
the key mix
236
00:11:27,400 --> 00:11:31,000
okay so I'm going to apply this on the sides
237
00:11:31,766 --> 00:11:34,399
then this on the other side
238
00:11:34,400 --> 00:11:37,000
because I want to have before the grain
239
00:11:37,000 --> 00:11:38,633
and after the grain
240
00:11:40,066 --> 00:11:43,533
okay so here you are
241
00:11:43,533 --> 00:11:45,366
and now the mask
242
00:11:45,366 --> 00:11:47,833
okay and this
243
00:11:49,200 --> 00:11:51,466
is the mask that is dealing with that
244
00:11:51,466 --> 00:11:55,266
so I'm going to invert the mask okay
245
00:11:55,266 --> 00:12:00,699
so now I'm applying the grain just through the alpha
246
00:12:00,800 --> 00:12:05,533
and why I'm spending that much with the alpha
247
00:12:05,533 --> 00:12:07,066
or the Primo duplication
248
00:12:07,066 --> 00:12:08,699
have a look what's happening
249
00:12:08,866 --> 00:12:14,066
when I'm not applying any kind of limit to the alpha
250
00:12:14,133 --> 00:12:16,599
or to the area of interest
251
00:12:16,666 --> 00:12:18,766
so if you go to the original
252
00:12:18,766 --> 00:12:20,366
and now everything
253
00:12:20,866 --> 00:12:23,299
the limiters have been deactivated
254
00:12:23,366 --> 00:12:26,299
if I rising here the gamma
255
00:12:26,300 --> 00:12:29,366
you will see that the grain has been double
256
00:12:30,733 --> 00:12:32,999
see and this is an error
257
00:12:33,000 --> 00:12:35,866
because you are touching the background plate
258
00:12:35,866 --> 00:12:36,533
and remember
259
00:12:36,533 --> 00:12:39,633
one of the golden rules is never to touch
260
00:12:39,666 --> 00:12:41,866
the background plate if you don't need it
261
00:12:41,866 --> 00:12:44,933
okay so this is adding just an issue
262
00:12:44,933 --> 00:12:48,099
so that's why we're going to use this
263
00:12:48,733 --> 00:12:49,999
that is going to limit
264
00:12:50,133 --> 00:12:55,266
using the alpha masking or the amp multiply
265
00:12:55,266 --> 00:12:58,266
okay and you will find that is quite similar
266
00:12:58,366 --> 00:12:59,499
the difference is
267
00:12:59,500 --> 00:13:03,333
you can play a beat with the density of the alpha
268
00:13:03,333 --> 00:13:06,699
if you are using the key mix option
269
00:13:06,766 --> 00:13:08,866
what I mean is in here
270
00:13:09,333 --> 00:13:11,599
I can place a great
271
00:13:12,866 --> 00:13:15,666
and I'm going to use just the alpha channel
272
00:13:15,700 --> 00:13:17,033
and then with the gamma
273
00:13:17,566 --> 00:13:19,599
I can increase or decrease
274
00:13:19,700 --> 00:13:21,766
and you will notice that in here
275
00:13:21,766 --> 00:13:24,966
in this area that is coming through the window
276
00:13:25,133 --> 00:13:28,366
you're going to apply more or less
277
00:13:29,533 --> 00:13:32,799
okay so the question is
278
00:13:32,800 --> 00:13:36,933
when you need to apply more or less grain
279
00:13:36,933 --> 00:13:42,099
through semi transparency or using the soft edges
280
00:13:42,266 --> 00:13:46,699
then you can use this Chemix option okay
281
00:13:47,100 --> 00:13:49,066
yes to apply the grain
282
00:13:49,066 --> 00:13:51,566
but this is the easiest example
283
00:13:51,566 --> 00:13:52,366
because of course
284
00:13:52,366 --> 00:13:54,699
in here we have the semi transparency
285
00:13:54,700 --> 00:13:57,400
already baked in the alpha
286
00:13:57,400 --> 00:14:00,566
but what happened when that
287
00:14:00,566 --> 00:14:04,333
semi transparency is not reflected in the alpha channel
288
00:14:04,333 --> 00:14:06,266
and is it possible
289
00:14:06,266 --> 00:14:07,366
of course it is
290
00:14:07,366 --> 00:14:11,499
because you can get volumetic light
291
00:14:11,700 --> 00:14:13,500
or incandin sense
292
00:14:13,500 --> 00:14:16,766
or kind of a glowy light from your CG
293
00:14:17,133 --> 00:14:20,766
and that is not going to be reflected in the alpha
294
00:14:20,766 --> 00:14:23,833
because that is going to be at
295
00:14:24,200 --> 00:14:26,300
to the resulting image at the end
296
00:14:26,300 --> 00:14:29,900
so remember that the other whatever is not in the
297
00:14:29,900 --> 00:14:33,733
in the alpha is just going to be like a plus operation
298
00:14:33,733 --> 00:14:37,599
okay so I prepare for you in here
299
00:14:38,866 --> 00:14:40,099
this option
300
00:14:40,100 --> 00:14:42,566
okay which is
301
00:14:42,566 --> 00:14:46,399
let me deactivate the set of the grain for a sec
302
00:14:46,400 --> 00:14:48,900
so you can see the result without the grain
303
00:14:48,933 --> 00:14:50,899
so what happens
304
00:14:50,900 --> 00:14:51,766
for instance
305
00:14:51,766 --> 00:14:55,466
if I get those lens flares in there
306
00:14:55,733 --> 00:14:58,666
so if you have a look at the CG
307
00:14:59,200 --> 00:15:01,333
the Alpha is exactly the same
308
00:15:01,333 --> 00:15:06,166
because those lens flares are not solid
309
00:15:06,166 --> 00:15:08,666
are just light to be at on top
310
00:15:08,666 --> 00:15:09,799
in this case
311
00:15:09,800 --> 00:15:11,600
for the sake of this example
312
00:15:11,600 --> 00:15:18,533
I yet the flares over the CG branch of my compositing
313
00:15:18,533 --> 00:15:21,466
you can add it at the end when everything is mixed
314
00:15:21,533 --> 00:15:24,333
but again you're going to have the same problem
315
00:15:24,333 --> 00:15:27,766
is those flares has no grain
316
00:15:27,800 --> 00:15:30,466
and that is quite unnatural okay
317
00:15:31,100 --> 00:15:36,666
so what if I apply the same trick as before
318
00:15:36,800 --> 00:15:41,500
so using the amp remote and then the grain
319
00:15:41,500 --> 00:15:43,033
and then the promote
320
00:15:43,466 --> 00:15:44,266
here you are
321
00:15:45,600 --> 00:15:52,400
you just kill your flares outside the Alpha okay
322
00:15:52,800 --> 00:15:54,200
and this is why
323
00:15:54,666 --> 00:15:58,499
because the alpha now on after the grain
324
00:15:58,866 --> 00:16:02,933
is going to crop everything that is in black in there
325
00:16:02,933 --> 00:16:06,866
and because your flares have no alpha
326
00:16:06,900 --> 00:16:09,133
they are going to be crop out
327
00:16:09,133 --> 00:16:13,099
so this is a narrow okay
328
00:16:13,166 --> 00:16:15,433
um you can also propose
329
00:16:15,700 --> 00:16:18,900
yeah why don't we use this trick
330
00:16:19,200 --> 00:16:21,700
the one of just as pleating the alpha
331
00:16:22,000 --> 00:16:23,500
then grading the alpha
332
00:16:23,666 --> 00:16:27,033
and then applying the mask for the alpha
333
00:16:27,133 --> 00:16:29,266
just to be dealing with the green
334
00:16:29,266 --> 00:16:30,366
well actually
335
00:16:30,366 --> 00:16:32,866
that is going to be exactly the same problem
336
00:16:32,866 --> 00:16:34,599
because the alpha
337
00:16:34,900 --> 00:16:39,233
he's not covering that area of the flares
338
00:16:39,266 --> 00:16:41,933
so you are going to apply the grain in here
339
00:16:41,933 --> 00:16:43,466
but nothing in there
340
00:16:43,466 --> 00:16:46,799
and probably is not going to be that visible okay
341
00:16:46,800 --> 00:16:50,266
so you are going to have the grain in the solid areas
342
00:16:50,266 --> 00:16:51,399
and no grain in there
343
00:16:51,400 --> 00:16:54,000
that at the end of the day is going to mix
344
00:16:54,566 --> 00:16:56,799
so if you see here
345
00:16:56,800 --> 00:16:58,466
this is exactly the same
346
00:16:59,200 --> 00:17:02,366
you are going to miss the grain in this area
347
00:17:02,700 --> 00:17:03,300
I'm probably
348
00:17:03,300 --> 00:17:06,800
is going to mix with the grain that is already in there
349
00:17:06,900 --> 00:17:09,166
but he's not correct
350
00:17:09,266 --> 00:17:12,499
because you are not adding the same amount
351
00:17:12,600 --> 00:17:16,433
of grain to the flare as is expected
352
00:17:16,533 --> 00:17:20,099
so we need to calculate a work around for that
353
00:17:20,400 --> 00:17:22,400
so for this instance
354
00:17:22,400 --> 00:17:28,000
what we need to get is away of extracting that effect
355
00:17:28,100 --> 00:17:29,833
just to be isolated
356
00:17:29,866 --> 00:17:32,699
and apply the grain just to the effect
357
00:17:32,700 --> 00:17:34,500
and then applying the grain
358
00:17:34,500 --> 00:17:37,200
just for the solid areas of the alpha
359
00:17:37,200 --> 00:17:41,766
okay so we're going to talk about effect separation
360
00:17:41,766 --> 00:17:44,866
okay so in here
361
00:17:45,000 --> 00:17:51,166
what I did is just to create a silly render in here
362
00:17:51,166 --> 00:17:52,799
just put a group
363
00:17:52,966 --> 00:17:56,166
just to make you think what to do
364
00:17:56,166 --> 00:17:58,599
when your barometric lights
365
00:17:58,600 --> 00:18:01,366
or your effects that are not in the alpha
366
00:18:01,500 --> 00:18:03,400
are baked in your render
367
00:18:03,400 --> 00:18:06,600
so if you want to grain that
368
00:18:06,700 --> 00:18:08,766
you need to figure out a way of
369
00:18:08,866 --> 00:18:12,299
getting the effect out of the render okay
370
00:18:13,266 --> 00:18:15,366
another thing to mention is
371
00:18:15,366 --> 00:18:16,666
if you have the
372
00:18:17,000 --> 00:18:21,066
the flares separate in here
373
00:18:21,066 --> 00:18:23,566
you can also propose me ah yeah
374
00:18:23,566 --> 00:18:27,799
why don't we apply the grain just to the flares
375
00:18:28,533 --> 00:18:32,166
and then mix everything together again
376
00:18:33,600 --> 00:18:36,766
well what is actually happening is
377
00:18:36,766 --> 00:18:40,799
you are going to add graining here for the car
378
00:18:40,800 --> 00:18:43,866
and then you're going to add more graining here
379
00:18:43,866 --> 00:18:44,766
for the lens flare
380
00:18:44,766 --> 00:18:46,766
so hmm yeah
381
00:18:46,766 --> 00:18:48,299
probably is gonna work
382
00:18:48,300 --> 00:18:51,900
but is not a good mathematically correct approach
383
00:18:51,900 --> 00:18:53,800
okay so remember
384
00:18:53,866 --> 00:18:54,499
in compositing
385
00:18:54,500 --> 00:18:55,533
everything is cheating
386
00:18:55,533 --> 00:18:58,066
so whatever is going to work is going to work
387
00:18:58,066 --> 00:19:00,066
so nothing to say about that
388
00:19:00,066 --> 00:19:01,899
but if you want to find
389
00:19:01,900 --> 00:19:05,233
a procedural way of getting to a good result
390
00:19:05,300 --> 00:19:08,166
this is how I'm going to show you
391
00:19:08,166 --> 00:19:13,299
okay so I'm going to revert to the flares in there
392
00:19:13,500 --> 00:19:15,633
and in here
393
00:19:15,866 --> 00:19:18,699
what I did is baked everything in one notes
394
00:19:18,700 --> 00:19:23,600
so how can I great the areas beyond the alpha
395
00:19:23,600 --> 00:19:25,400
well in here
396
00:19:25,500 --> 00:19:30,466
what I did is just to create the same effect that the
397
00:19:30,533 --> 00:19:34,133
promote application is going to create with the grain
398
00:19:34,133 --> 00:19:36,499
okay so in here
399
00:19:36,933 --> 00:19:41,766
this is my car with the effect of the amprimote
400
00:19:41,766 --> 00:19:42,966
and then the primote
401
00:19:42,966 --> 00:19:46,033
so I'm just getting rid of the effect
402
00:19:46,333 --> 00:19:49,466
so what if I have this
403
00:19:50,066 --> 00:19:51,466
and I subtract
404
00:19:52,966 --> 00:19:56,899
this so this is what I'm doing actually with this merch
405
00:19:56,900 --> 00:19:59,500
this is an effect extraction
406
00:19:59,500 --> 00:20:02,200
you can use it for many other things okay
407
00:20:02,200 --> 00:20:05,266
so you have in here the minor
408
00:20:05,266 --> 00:20:07,466
so the original image
409
00:20:07,800 --> 00:20:11,033
minus the one with the effect we want to isolate
410
00:20:11,100 --> 00:20:14,600
and here we have the isolation
411
00:20:14,766 --> 00:20:17,833
okay and you can apply this even for the
412
00:20:17,933 --> 00:20:19,533
for this okay
413
00:20:19,533 --> 00:20:21,266
so if you have
414
00:20:23,600 --> 00:20:25,800
this minus that
415
00:20:29,133 --> 00:20:32,066
what you are going to get is just the flares okay
416
00:20:32,066 --> 00:20:36,599
so in case you have effects that are baked in the image
417
00:20:36,600 --> 00:20:38,500
or you need the image to process
418
00:20:38,500 --> 00:20:39,100
for instance
419
00:20:39,100 --> 00:20:41,300
to create a filter kick
420
00:20:41,300 --> 00:20:44,166
or a chromatic aberration or a nail
421
00:20:44,166 --> 00:20:48,633
or any other effects that requires the original image
422
00:20:48,666 --> 00:20:52,699
the way of extrapolating that effect after the process
423
00:20:52,700 --> 00:20:54,133
is using just this
424
00:20:54,133 --> 00:20:55,899
and then how do you mix it
425
00:20:55,900 --> 00:20:57,566
well it's really easy
426
00:20:57,900 --> 00:21:00,866
you put a minus to isolate
427
00:21:01,066 --> 00:21:05,499
then you put a blast to integrate
428
00:21:05,866 --> 00:21:07,099
okay but remember
429
00:21:07,100 --> 00:21:09,400
needs to be done with the original okay
430
00:21:09,700 --> 00:21:14,566
so that is the method for isolating the effect
431
00:21:14,733 --> 00:21:17,166
so in here we got that
432
00:21:17,166 --> 00:21:18,399
so we are going to have
433
00:21:18,400 --> 00:21:20,666
two different processes of the grain
434
00:21:20,900 --> 00:21:24,100
one for the areas of the alpha in here
435
00:21:24,133 --> 00:21:25,766
that is having this grain
436
00:21:26,366 --> 00:21:27,399
and the other
437
00:21:29,200 --> 00:21:32,566
for the areas that are not cover by the alpha
438
00:21:32,666 --> 00:21:36,633
so here what I'm doing is I'm going to apply the grain
439
00:21:36,800 --> 00:21:40,366
but remember that if I apply the grain to everything
440
00:21:40,866 --> 00:21:41,999
I'm going to have grain
441
00:21:42,000 --> 00:21:44,900
even in the areas I don't want the grain to be apply
442
00:21:44,900 --> 00:21:47,266
um because in here there is no alpha
443
00:21:47,500 --> 00:21:49,666
you need to invent the alpha yourself
444
00:21:49,666 --> 00:21:53,266
so if you are using just light
445
00:21:53,266 --> 00:21:54,799
um in this case
446
00:21:54,966 --> 00:21:57,399
the lens flare is just light
447
00:21:57,500 --> 00:22:00,366
what you should do is use aluminum ski
448
00:22:00,366 --> 00:22:02,799
so it's going to be base on the light
449
00:22:02,933 --> 00:22:04,933
so I'm in here
450
00:22:04,933 --> 00:22:07,966
just mixing the original without any grain
451
00:22:07,966 --> 00:22:10,733
and then the grain overall
452
00:22:10,733 --> 00:22:11,899
okay so now
453
00:22:11,900 --> 00:22:15,900
I'm going to deal with it using this luminance key
454
00:22:15,966 --> 00:22:19,066
okay so the luminance key here is just
455
00:22:19,266 --> 00:22:24,433
reflecting all the areas that are being touched by that
456
00:22:25,933 --> 00:22:26,999
luminance okay
457
00:22:27,000 --> 00:22:29,700
so here now
458
00:22:30,400 --> 00:22:31,566
as you can see
459
00:22:32,100 --> 00:22:35,100
I'm applying the grain on those areas
460
00:22:35,100 --> 00:22:36,866
but nothing beyond
461
00:22:36,866 --> 00:22:39,666
okay this is the difference okay
462
00:22:39,666 --> 00:22:42,533
so you can see that there is a tiny Halo
463
00:22:42,533 --> 00:22:46,166
where the green is just decading
464
00:22:46,166 --> 00:22:48,199
so is not going to be at full
465
00:22:48,200 --> 00:22:49,700
well that is normal
466
00:22:49,700 --> 00:22:51,966
and then you need to deal with that
467
00:22:52,100 --> 00:22:54,766
when you get the final image
468
00:22:54,766 --> 00:22:56,666
so here what I'm doing is
469
00:22:56,766 --> 00:23:01,466
I have the green with the premotified areas
470
00:23:01,466 --> 00:23:06,733
then the grain of the non premotified areas altogether
471
00:23:06,733 --> 00:23:08,599
using a plus okay
472
00:23:08,733 --> 00:23:12,266
and then in here you have altogether
473
00:23:12,300 --> 00:23:15,000
and not is that in here you have the grain
474
00:23:15,166 --> 00:23:19,899
see that is going to be a very tiny variation
475
00:23:20,133 --> 00:23:21,366
base is still there
476
00:23:21,500 --> 00:23:23,300
and that is important
477
00:23:23,366 --> 00:23:25,899
so if you want to change that
478
00:23:26,100 --> 00:23:28,966
you can change in the luminance key
479
00:23:29,966 --> 00:23:34,033
the amptitude of that area of the alpha
480
00:23:34,466 --> 00:23:37,266
the problem is the keyer
481
00:23:37,300 --> 00:23:39,600
the note has the limitation
482
00:23:39,600 --> 00:23:44,433
of just setting a black point and a white point
483
00:23:44,566 --> 00:23:47,499
but there is nothing to work in the middle
484
00:23:47,600 --> 00:23:49,900
what is called fall off
485
00:23:49,933 --> 00:23:51,966
or most simply
486
00:23:51,966 --> 00:23:54,766
the gamma between this and that point
487
00:23:54,766 --> 00:23:59,699
so I try not to use the luminance key much
488
00:23:59,700 --> 00:24:02,233
I mean the cure for the luminance key
489
00:24:02,333 --> 00:24:07,133
and I prefer to use my own luminance key setup
490
00:24:07,133 --> 00:24:11,799
so what I did in here is a variation of this
491
00:24:11,966 --> 00:24:16,566
where I can not only use the black and the white point
492
00:24:16,566 --> 00:24:19,866
but also use the gamma here
493
00:24:20,333 --> 00:24:22,966
as a feather fell off
494
00:24:22,966 --> 00:24:25,066
so I'm going to show you how I did it
495
00:24:25,066 --> 00:24:27,466
so this is my original image
496
00:24:28,566 --> 00:24:32,099
what I did is apply a saturation note okay
497
00:24:32,100 --> 00:24:35,500
you can find it in here saturation
498
00:24:35,866 --> 00:24:38,399
and I reduce the saturation to zero
499
00:24:38,400 --> 00:24:41,900
using the luminance math wreck 7:09
500
00:24:41,966 --> 00:24:45,466
which means what I'm going to get after this saturation
501
00:24:45,466 --> 00:24:48,533
is just the luminance of this image
502
00:24:48,533 --> 00:24:51,999
according to the standard of the Rex 7:09
503
00:24:52,000 --> 00:24:53,166
which is by the way
504
00:24:53,166 --> 00:24:56,966
the same standard that is using the luminance key okay
505
00:24:56,966 --> 00:25:03,233
so this is just creating the luminance map of my image
506
00:25:03,266 --> 00:25:05,866
then what I'm doing here with this grade
507
00:25:05,866 --> 00:25:08,933
is applying the same controls I have here
508
00:25:08,933 --> 00:25:12,266
so this point that is now at 0 0
509
00:25:12,266 --> 00:25:14,966
0 0 is the black point
510
00:25:15,000 --> 00:25:18,133
this here on top is the white point
511
00:25:18,133 --> 00:25:19,999
okay so what I did
512
00:25:20,100 --> 00:25:23,900
is just use the black point and the white point
513
00:25:23,900 --> 00:25:29,266
so I can set the amplitude of that luminance key
514
00:25:29,500 --> 00:25:30,866
but in here
515
00:25:30,866 --> 00:25:31,666
the difference on
516
00:25:31,666 --> 00:25:33,599
for me is a massive difference
517
00:25:33,600 --> 00:25:37,366
is you can study the decay of that mat
518
00:25:37,366 --> 00:25:39,566
so you can move in here
519
00:25:41,066 --> 00:25:42,666
and you can distribute
520
00:25:43,133 --> 00:25:46,533
how you want that gamma to appear
521
00:25:46,533 --> 00:25:47,933
the feather fall off
522
00:25:47,933 --> 00:25:48,699
for instance to
523
00:25:48,700 --> 00:25:51,466
to create a correlation with the roton out
524
00:25:51,466 --> 00:25:53,666
okay so in here
525
00:25:53,666 --> 00:25:55,466
if I apply that
526
00:25:56,166 --> 00:25:58,566
you're gonna notice very
527
00:25:58,566 --> 00:26:02,699
very tiny variation on how the grain is going to be
528
00:26:02,700 --> 00:26:06,266
apply through the semi transparent areas
529
00:26:06,266 --> 00:26:07,333
but even more
530
00:26:07,333 --> 00:26:11,499
you can just play with the white and black point
531
00:26:11,700 --> 00:26:15,400
and just to isolate that area and apply
532
00:26:15,666 --> 00:26:19,533
way more or less grain in the areas of your interest
533
00:26:19,533 --> 00:26:20,366
in this case
534
00:26:20,366 --> 00:26:22,333
is not going to be that visible
535
00:26:22,333 --> 00:26:23,599
but remember this
536
00:26:23,600 --> 00:26:25,200
when you are going to have
537
00:26:25,500 --> 00:26:29,766
quite contrast the areas in luminance terms okay
538
00:26:29,766 --> 00:26:31,499
something to keep in mind
539
00:26:31,500 --> 00:26:33,700
when you are creating a luminance key
540
00:26:33,700 --> 00:26:36,233
is every time you are working with a mat
541
00:26:36,700 --> 00:26:38,566
remember that you want
542
00:26:38,800 --> 00:26:42,600
all your values to be normalized between 0 and 1
543
00:26:42,600 --> 00:26:44,566
you don't want values below 0
544
00:26:44,566 --> 00:26:45,533
so negative
545
00:26:45,533 --> 00:26:49,066
or you don't want values to be super white above 1
546
00:26:49,066 --> 00:26:53,466
so every time you're creating this luminance key
547
00:26:53,466 --> 00:26:54,666
in a manual way
548
00:26:54,666 --> 00:26:56,599
remember that in the great
549
00:26:56,733 --> 00:27:01,966
you should take the black and the white clamp on okay
550
00:27:01,966 --> 00:27:03,299
if you forget that
551
00:27:03,300 --> 00:27:06,700
and then you increase the white point sometimes
552
00:27:06,700 --> 00:27:09,900
you are going to increase your values beyond 1
553
00:27:09,900 --> 00:27:13,366
and that is going to create funky effects in the Chemix
554
00:27:13,366 --> 00:27:16,899
okay so just that to keep in mind
555
00:27:16,900 --> 00:27:20,666
so this could be a good effect to apply
556
00:27:20,733 --> 00:27:24,833
for the areas that are not in the alpha okay
557
00:27:25,533 --> 00:27:32,199
but what if the proling is actually the alpha well here
558
00:27:33,500 --> 00:27:36,433
what I did is just to apply a patch
559
00:27:37,200 --> 00:27:40,633
and the patch is going to be perfectly grain
560
00:27:40,700 --> 00:27:42,133
using the every grain
561
00:27:42,133 --> 00:27:46,233
so the structure of the grain is perfectly done
562
00:27:46,400 --> 00:27:49,066
but here you're gonna notice that
563
00:27:49,366 --> 00:27:51,766
between this area of the grain
564
00:27:52,133 --> 00:27:53,999
and this area of the grain
565
00:27:54,133 --> 00:27:57,499
there is some kind of softening
566
00:27:57,500 --> 00:28:01,200
here there is less grain than I expected
567
00:28:01,500 --> 00:28:04,166
and this is because I'm mixing
568
00:28:04,166 --> 00:28:06,166
the grain of the original plate
569
00:28:06,166 --> 00:28:08,399
with the grain of my patch
570
00:28:08,500 --> 00:28:11,900
and due to the graviations in here
571
00:28:12,200 --> 00:28:13,866
that in here
572
00:28:13,866 --> 00:28:15,233
around the center
573
00:28:15,600 --> 00:28:18,966
is creating that blending effect
574
00:28:18,966 --> 00:28:20,433
I show you before
575
00:28:21,166 --> 00:28:24,066
with the averaging of the grains
576
00:28:24,066 --> 00:28:28,666
so is actually contrasting each grain part
577
00:28:28,733 --> 00:28:32,233
so when you are adding two grains
578
00:28:32,266 --> 00:28:34,166
the result is less grain
579
00:28:34,566 --> 00:28:36,466
so how to fix that
580
00:28:36,600 --> 00:28:41,533
well let me show you in here a few ways of figuring out
581
00:28:41,533 --> 00:28:44,266
because there is no one single way
582
00:28:44,500 --> 00:28:47,500
depending on where are you applying the grain
583
00:28:47,500 --> 00:28:49,966
or your personal taste
584
00:28:50,000 --> 00:28:52,633
you're going to use one or another
585
00:28:52,766 --> 00:28:56,466
so let's start by this setup
586
00:28:56,566 --> 00:29:00,233
which is an special treatment of the fringe well
587
00:29:00,300 --> 00:29:01,566
to understand this
588
00:29:01,566 --> 00:29:03,399
you first need to understand
589
00:29:03,400 --> 00:29:04,800
what's the fringe
590
00:29:05,100 --> 00:29:08,833
the fringe represent in a mat
591
00:29:09,300 --> 00:29:13,066
whatever is not white or black
592
00:29:13,400 --> 00:29:19,666
so is increasing towards the center of the graviation
593
00:29:19,666 --> 00:29:23,633
between the inside and the outside
594
00:29:23,700 --> 00:29:27,500
okay so if I show you with an example
595
00:29:27,500 --> 00:29:29,966
is going to be easier to visualize
596
00:29:30,000 --> 00:29:33,166
so this is my shape
597
00:29:33,166 --> 00:29:36,199
which is just a circle with a soft area
598
00:29:36,666 --> 00:29:38,999
so in here the center of the circle
599
00:29:39,000 --> 00:29:42,033
is going to be the heart part of the circle
600
00:29:42,333 --> 00:29:45,766
in here the black is going to represent the outside
601
00:29:46,300 --> 00:29:50,300
and if I apply a mask okay
602
00:29:50,400 --> 00:29:53,000
just to apply a color correction
603
00:29:53,000 --> 00:29:53,800
like in here
604
00:29:53,800 --> 00:29:56,366
a grade to create this red
605
00:29:57,100 --> 00:29:59,966
this is a regular mask okay
606
00:29:59,966 --> 00:30:06,833
so what if I want to use the fringe of this mat
607
00:30:07,266 --> 00:30:08,899
to apply the color correction
608
00:30:08,900 --> 00:30:11,900
so this is the fringe in here
609
00:30:11,900 --> 00:30:13,400
how I apply the fringe
610
00:30:13,400 --> 00:30:19,000
well I went to the properties of the masking here
611
00:30:19,000 --> 00:30:20,466
of the color correction
612
00:30:20,533 --> 00:30:23,633
and you will find that this setup of the mask
613
00:30:23,966 --> 00:30:25,766
and inject invert
614
00:30:25,866 --> 00:30:28,099
and the fringe is available
615
00:30:28,100 --> 00:30:31,066
and you can see here that is actually
616
00:30:31,066 --> 00:30:32,566
using the pictures
617
00:30:32,566 --> 00:30:34,166
that are not outside
618
00:30:34,266 --> 00:30:36,966
not inside but in the middle
619
00:30:37,000 --> 00:30:39,766
and the more you are going towards
620
00:30:39,966 --> 00:30:42,466
the middle point of the gradient
621
00:30:42,566 --> 00:30:44,999
the more strength is going to have
622
00:30:45,000 --> 00:30:47,000
actually at the middle point
623
00:30:47,000 --> 00:30:48,566
is going to be full
624
00:30:48,566 --> 00:30:51,499
and then is going to decay towards the inside
625
00:30:51,500 --> 00:30:53,300
and towards the outside
626
00:30:53,300 --> 00:30:56,000
let me show you with the graphical representation
627
00:30:56,000 --> 00:30:57,866
so if this is the ramp
628
00:30:57,866 --> 00:31:00,366
as we saw in the slice
629
00:31:00,366 --> 00:31:02,499
to remember in here
630
00:31:02,733 --> 00:31:05,466
this is the normal ramp
631
00:31:05,733 --> 00:31:09,666
so the middle point is in here
632
00:31:09,733 --> 00:31:11,799
so in that middle point
633
00:31:11,966 --> 00:31:14,733
the strength of the mask is going to be at full
634
00:31:14,733 --> 00:31:16,066
so it's going to be here
635
00:31:16,066 --> 00:31:17,799
let me put it in another color
636
00:31:17,800 --> 00:31:21,833
so the strength is going to decay from the middle point
637
00:31:22,200 --> 00:31:24,200
to the outside
638
00:31:24,933 --> 00:31:26,499
and to the inside
639
00:31:26,500 --> 00:31:28,400
so it's actually doing this
640
00:31:31,900 --> 00:31:36,100
okay so is converting this linear ramp
641
00:31:36,133 --> 00:31:38,666
to this kind of triangle
642
00:31:38,666 --> 00:31:40,799
or pyramite okay
643
00:31:41,166 --> 00:31:44,699
so in here is the maximum strength
644
00:31:45,500 --> 00:31:48,133
this curve that I'm representing here
645
00:31:48,133 --> 00:31:51,566
this triangle is called solarisation
646
00:31:51,966 --> 00:31:55,766
okay so the solarization operation is actually
647
00:31:55,766 --> 00:31:57,733
to take those values
648
00:31:57,733 --> 00:32:00,333
and to create the maximum strength
649
00:32:00,333 --> 00:32:01,733
towards the middle point
650
00:32:01,733 --> 00:32:05,699
and then decay towards the 1 or the 0 values
651
00:32:05,700 --> 00:32:08,266
okay so it's going to be easier
652
00:32:09,133 --> 00:32:12,799
if I show you with a color look up
653
00:32:12,900 --> 00:32:14,800
okay so this is my image
654
00:32:14,800 --> 00:32:17,766
and this is the solarization operation
655
00:32:17,766 --> 00:32:19,166
using a color look up
656
00:32:19,166 --> 00:32:20,466
so a color look up
657
00:32:20,466 --> 00:32:21,133
as you know
658
00:32:21,133 --> 00:32:23,099
is just the curves
659
00:32:23,200 --> 00:32:27,633
okay so this is a solarization
660
00:32:27,900 --> 00:32:31,766
with the horizontal interpolation at the peak
661
00:32:31,766 --> 00:32:34,199
which means just that at this point
662
00:32:34,200 --> 00:32:37,800
and just using this softening for the decay
663
00:32:37,800 --> 00:32:40,066
okay but I can change that
664
00:32:40,266 --> 00:32:43,099
and use a more strict solarization
665
00:32:43,100 --> 00:32:45,666
by using the linear interpolation
666
00:32:45,666 --> 00:32:49,333
so this is an extreme solarization
667
00:32:49,333 --> 00:32:52,466
okay so the middle point is the cane
668
00:32:52,700 --> 00:32:54,166
in a linear fashion
669
00:32:54,166 --> 00:32:56,533
towards 0 and towards 1
670
00:32:56,533 --> 00:32:57,599
and you can see
671
00:32:57,766 --> 00:33:02,499
the only part or portion of the image that is taking
672
00:33:02,500 --> 00:33:06,400
is just what is not white or black
673
00:33:06,500 --> 00:33:10,766
okay so this is the solarization
674
00:33:10,766 --> 00:33:12,699
using a color look up okay
675
00:33:12,700 --> 00:33:15,900
so if you want to create a solarization by yourself
676
00:33:15,900 --> 00:33:19,033
is just as simple as taking
677
00:33:20,366 --> 00:33:22,566
a color look up here
678
00:33:24,400 --> 00:33:27,366
and just to go to the master or the alpha
679
00:33:27,366 --> 00:33:29,166
depending on the channel you want
680
00:33:29,200 --> 00:33:33,100
you create by pressing Alt and command in the centre
681
00:33:33,100 --> 00:33:34,300
then in here
682
00:33:34,300 --> 00:33:38,933
you modify the coordinates by manual inputs
683
00:33:38,933 --> 00:33:41,666
and then the X must be the same
684
00:33:41,766 --> 00:33:44,866
and in here the y
685
00:33:44,900 --> 00:33:46,066
as I put it
686
00:33:47,266 --> 00:33:50,066
the y must be 0 okay
687
00:33:50,066 --> 00:33:51,299
and then in the middle
688
00:33:51,366 --> 00:33:54,499
this must be point five
689
00:33:56,066 --> 00:33:59,899
and the Y1 okay
690
00:34:00,966 --> 00:34:04,566
and this is a soft solarize
691
00:34:04,733 --> 00:34:07,899
if you want the strict solarized
692
00:34:07,900 --> 00:34:10,400
you go to the interpolation linear
693
00:34:10,600 --> 00:34:13,633
and that is the solarization okay
694
00:34:14,066 --> 00:34:16,566
so you can create by yourself
695
00:34:16,566 --> 00:34:17,599
and then you can decide
696
00:34:17,600 --> 00:34:21,733
if you want more portions towards the outside
697
00:34:21,733 --> 00:34:24,533
or more portions towards the inside
698
00:34:24,533 --> 00:34:26,333
then you can play okay
699
00:34:26,333 --> 00:34:31,799
so I find myself easier to work with the color look up
700
00:34:31,800 --> 00:34:36,033
because sometimes you want to create a portion manually
701
00:34:36,166 --> 00:34:38,566
or just reduce the other side
702
00:34:38,866 --> 00:34:42,966
and in here you can do it graphically okay
703
00:34:43,466 --> 00:34:44,133
by contrary
704
00:34:44,133 --> 00:34:46,766
if you are using the automatic way
705
00:34:46,766 --> 00:34:49,166
that is the only setup you have
706
00:34:49,166 --> 00:34:51,566
okay so this is just for you
707
00:34:51,566 --> 00:34:55,033
to let you know that you can modify that mask
708
00:34:55,066 --> 00:34:57,666
whatever is going to work for your image
709
00:34:57,666 --> 00:34:58,833
is going to be good
710
00:34:58,866 --> 00:35:01,133
but at least you need to know how it works
711
00:35:01,133 --> 00:35:06,566
okay so let's go back to the result image
712
00:35:06,566 --> 00:35:08,299
because what I did in here
713
00:35:08,300 --> 00:35:10,933
is quite similar to the previous example
714
00:35:10,933 --> 00:35:11,999
but in this instance
715
00:35:12,000 --> 00:35:13,466
I change the method
716
00:35:13,466 --> 00:35:14,966
so what I did is
717
00:35:14,966 --> 00:35:19,199
I'm going to apply the noise for everything
718
00:35:19,200 --> 00:35:21,700
of course every time I'm applying a patch
719
00:35:21,700 --> 00:35:23,066
I'm going to the noise
720
00:35:23,066 --> 00:35:25,899
because I don't want to double the the grain
721
00:35:25,966 --> 00:35:27,999
I'm going to apply my patch
722
00:35:28,766 --> 00:35:31,566
and that patch is going to be pre
723
00:35:31,566 --> 00:35:34,966
multiply by this shape
724
00:35:35,400 --> 00:35:37,666
so in here I'm applying the alpha
725
00:35:38,133 --> 00:35:40,066
and I'm going to pre multiply
726
00:35:40,566 --> 00:35:44,099
then I'm going to put everything over okay
727
00:35:44,333 --> 00:35:48,999
and then what I'm going to do is an extra pass of grain
728
00:35:49,100 --> 00:35:51,966
and remember what I mentioned before about the sale
729
00:35:51,966 --> 00:35:53,999
randomization of the values
730
00:35:54,066 --> 00:35:55,966
what I mean is in here
731
00:35:56,133 --> 00:35:59,999
I need to apply exactly the same grain
732
00:36:00,000 --> 00:36:01,900
I'm applying to the original patch
733
00:36:01,900 --> 00:36:05,133
I cannot apply another layer of grain
734
00:36:05,133 --> 00:36:06,066
completely different
735
00:36:06,066 --> 00:36:09,766
because that the result is going to lose more grain
736
00:36:09,766 --> 00:36:10,999
that I'm going to add
737
00:36:11,133 --> 00:36:12,666
so what I did is
738
00:36:12,666 --> 00:36:14,933
I clone this grain
739
00:36:14,933 --> 00:36:16,366
and this other grain
740
00:36:16,366 --> 00:36:20,066
so the pattern and the randomization seat
741
00:36:20,066 --> 00:36:21,733
that I'm getting from each
742
00:36:21,733 --> 00:36:23,466
is exactly the same
743
00:36:23,466 --> 00:36:27,033
so in here I have the same setup
744
00:36:27,333 --> 00:36:29,099
that I got in the beginning
745
00:36:29,100 --> 00:36:32,600
but the point that I'm adding in here is
746
00:36:32,600 --> 00:36:36,433
I'm adding more grain to everything okay
747
00:36:37,200 --> 00:36:38,166
and in here
748
00:36:38,200 --> 00:36:40,066
and using the fringe
749
00:36:40,966 --> 00:36:43,699
just to isolate that area that is
750
00:36:43,700 --> 00:36:45,233
that is getting soft
751
00:36:45,900 --> 00:36:48,400
and I'm adding that extra grain
752
00:36:48,533 --> 00:36:51,733
so here have a look at difference
753
00:36:51,733 --> 00:36:53,999
so I'm going to put in there
754
00:36:54,000 --> 00:36:56,833
see that here
755
00:36:57,333 --> 00:36:58,399
you can notice
756
00:36:58,666 --> 00:37:00,966
it's not a massive difference in this case
757
00:37:00,966 --> 00:37:03,866
but you will find that in some shots
758
00:37:03,933 --> 00:37:06,699
this difference is going to be quite a lot
759
00:37:06,866 --> 00:37:09,099
okay so with this
760
00:37:09,100 --> 00:37:10,566
the nice thing is
761
00:37:10,566 --> 00:37:13,066
I can play with the key mix
762
00:37:13,200 --> 00:37:15,433
and going towards
763
00:37:15,700 --> 00:37:17,500
a more grainy results
764
00:37:17,500 --> 00:37:18,700
or reducing
765
00:37:19,600 --> 00:37:21,200
to a natural
766
00:37:21,200 --> 00:37:24,333
without any grain add
767
00:37:24,333 --> 00:37:25,733
okay so in here
768
00:37:25,733 --> 00:37:27,066
you can play a bit
769
00:37:27,066 --> 00:37:29,399
I find that usually
770
00:37:29,466 --> 00:37:33,099
the best result is around 0.25
771
00:37:33,100 --> 00:37:34,300
because of course
772
00:37:34,300 --> 00:37:37,233
you are getting already grain in there
773
00:37:37,333 --> 00:37:39,299
and it's mixing with the original grain
774
00:37:39,300 --> 00:37:42,333
so he's going to need kind of a
775
00:37:42,333 --> 00:37:44,399
an extra quarter just
776
00:37:44,400 --> 00:37:46,600
just for making the maths to work
777
00:37:46,600 --> 00:37:48,166
okay but it's not a rule
778
00:37:48,166 --> 00:37:50,766
so you will find that sometimes you need to add more
779
00:37:50,766 --> 00:37:52,899
sometimes you need to add less
780
00:37:52,966 --> 00:37:54,499
and at the end of the day
781
00:37:54,500 --> 00:37:57,700
this is an eye by eye process okay
782
00:37:58,200 --> 00:38:00,933
so in this case
783
00:38:00,933 --> 00:38:03,399
we find a work around
784
00:38:03,466 --> 00:38:07,599
to get the grain in those soft areas
785
00:38:08,466 --> 00:38:10,599
let's have a look at another way
786
00:38:10,766 --> 00:38:13,499
which is adding rain to the alpha
787
00:38:13,900 --> 00:38:16,066
well in this case
788
00:38:16,900 --> 00:38:20,633
probably this is not gonna work that much
789
00:38:20,666 --> 00:38:22,366
because I'm going to find
790
00:38:22,666 --> 00:38:23,766
like in here
791
00:38:23,766 --> 00:38:25,866
that the patch is not perfectly
792
00:38:26,366 --> 00:38:28,199
blend with the
793
00:38:28,366 --> 00:38:29,699
softening effect
794
00:38:30,066 --> 00:38:31,933
but this is just a good example
795
00:38:31,933 --> 00:38:33,633
to let you know that
796
00:38:33,933 --> 00:38:35,499
not all the techniques
797
00:38:35,500 --> 00:38:37,200
are going to work the same
798
00:38:37,200 --> 00:38:38,600
on all situations
799
00:38:38,600 --> 00:38:41,533
so that's why I want you to know all of them
800
00:38:41,533 --> 00:38:44,733
so you decide what to use in each case
801
00:38:44,733 --> 00:38:46,699
okay so when you need it
802
00:38:47,166 --> 00:38:48,966
what I did in here
803
00:38:49,533 --> 00:38:51,999
is actually blend
804
00:38:54,166 --> 00:39:00,766
the softening of my mat with differ
805
00:39:01,100 --> 00:39:02,533
differ is again
806
00:39:02,533 --> 00:39:05,866
a subtle randomization of frequencies in here
807
00:39:05,866 --> 00:39:11,666
so it's just kind of grain that I create in mono
808
00:39:11,666 --> 00:39:14,533
I means there is no randomization between the channels
809
00:39:14,533 --> 00:39:15,999
because what I'm going to do
810
00:39:16,000 --> 00:39:18,366
is just to apply it through the alpha
811
00:39:18,366 --> 00:39:19,799
so the alpha is just one channel
812
00:39:19,800 --> 00:39:22,833
so he's not going to make any difference actually
813
00:39:22,966 --> 00:39:25,266
so remember that in the beginning
814
00:39:25,266 --> 00:39:28,299
I was telling you about the frequencies
815
00:39:28,366 --> 00:39:33,099
so the grain in here
816
00:39:33,600 --> 00:39:35,100
the bigger is the grain
817
00:39:35,766 --> 00:39:38,533
the less frequency you are going to get
818
00:39:38,533 --> 00:39:41,299
in your grain pattern that you just created
819
00:39:41,366 --> 00:39:42,666
the differ by default
820
00:39:42,666 --> 00:39:45,966
is going to have a very high frequency
821
00:39:45,966 --> 00:39:48,933
because it's randomizing from pixel to pixels
822
00:39:48,933 --> 00:39:51,566
so if you apply a blur
823
00:39:51,800 --> 00:39:55,100
you will see that the effect you are going to observe
824
00:39:55,100 --> 00:39:58,766
is like the grain is actually being bigger
825
00:39:58,966 --> 00:39:59,499
but actually
826
00:39:59,500 --> 00:40:02,966
what you are doing is just creating clusters of pixels
827
00:40:02,966 --> 00:40:04,899
so the effect is actually
828
00:40:04,900 --> 00:40:06,500
that the grain is getting bigger
829
00:40:06,500 --> 00:40:09,400
but you are actually just changing the frequencies
830
00:40:09,400 --> 00:40:10,466
of that grain
831
00:40:10,500 --> 00:40:14,800
okay so I create this grain plate okay
832
00:40:15,900 --> 00:40:22,700
and what I did is to create it over a constant of 0.5
833
00:40:23,000 --> 00:40:26,233
because that is going to allocate
834
00:40:26,400 --> 00:40:28,766
all the randomizations
835
00:40:29,100 --> 00:40:32,766
starting from the middle point of the image okay
836
00:40:33,166 --> 00:40:37,899
so if I have a look at the slice tool
837
00:40:39,133 --> 00:40:40,199
in that point
838
00:40:42,166 --> 00:40:46,933
you will see that this is just constant image
839
00:40:46,933 --> 00:40:48,666
and as soon as I start with that
840
00:40:48,733 --> 00:40:52,566
see that is randomizing everything up and down
841
00:40:52,566 --> 00:40:53,699
quite strongly
842
00:40:53,700 --> 00:40:57,266
okay um in here
843
00:40:58,166 --> 00:40:59,199
I reduce the
844
00:40:59,200 --> 00:40:59,866
the frequency
845
00:40:59,866 --> 00:41:02,799
but this is still the amplitude
846
00:41:02,866 --> 00:41:04,566
is there okay
847
00:41:05,100 --> 00:41:07,066
now with this roto
848
00:41:07,066 --> 00:41:08,666
I'm applying an overly
849
00:41:08,666 --> 00:41:09,266
and remember
850
00:41:09,266 --> 00:41:12,233
that the overly is just adding the
851
00:41:12,700 --> 00:41:15,866
differences from the 0.5
852
00:41:15,866 --> 00:41:20,099
at we discuss in the class of the merge
853
00:41:20,200 --> 00:41:22,333
so what I'm going to do is
854
00:41:22,333 --> 00:41:25,899
I'm going to add the randomization of this grain
855
00:41:26,500 --> 00:41:28,166
to the areas in here
856
00:41:28,400 --> 00:41:29,633
not in the black
857
00:41:29,666 --> 00:41:30,799
not in the white
858
00:41:30,800 --> 00:41:32,300
but in the middle point
859
00:41:32,366 --> 00:41:36,866
so this is the effect of the overlay
860
00:41:37,000 --> 00:41:38,166
as you can see
861
00:41:38,166 --> 00:41:43,466
that is creating a grainy area of decay
862
00:41:43,766 --> 00:41:48,133
but is not perfectly white and black
863
00:41:48,133 --> 00:41:51,566
so you are having some areas that you are getting
864
00:41:51,566 --> 00:41:53,499
like a softness still
865
00:41:53,500 --> 00:41:54,966
same with the dark
866
00:41:54,966 --> 00:41:59,866
okay um that could be not noticeable
867
00:41:59,900 --> 00:42:03,333
but I prefer just to use a mathematical approach
868
00:42:03,333 --> 00:42:06,533
so if you don't want your grain to be notice
869
00:42:06,533 --> 00:42:09,933
remember that you need to have a hard edge
870
00:42:09,933 --> 00:42:12,099
so even if this is a grain
871
00:42:12,100 --> 00:42:14,033
the grain should be hard
872
00:42:14,133 --> 00:42:15,933
so what I did is
873
00:42:15,933 --> 00:42:18,533
I called the roll of contrast
874
00:42:18,533 --> 00:42:20,999
the roll of contrast is in here
875
00:42:21,333 --> 00:42:23,633
and this is a note
876
00:42:23,766 --> 00:42:26,199
where the contrast is going to be
877
00:42:26,200 --> 00:42:29,066
treat as mathematical value
878
00:42:29,066 --> 00:42:31,466
so is not like in the color correct
879
00:42:31,466 --> 00:42:36,266
this contrast is actually in a perceptible level
880
00:42:36,266 --> 00:42:41,799
okay so he's actually changing everything around 0.2
881
00:42:41,800 --> 00:42:44,800
which is the center of the middle point
882
00:42:45,133 --> 00:42:46,766
of the visual system in
883
00:42:46,766 --> 00:42:48,266
in the eye I mean
884
00:42:48,400 --> 00:42:49,533
but in here
885
00:42:49,533 --> 00:42:53,766
you decide where is the center of your transformations
886
00:42:53,766 --> 00:42:56,266
so in the center here
887
00:42:56,733 --> 00:43:00,866
what I did is just to increase the contrast quite high
888
00:43:00,866 --> 00:43:02,333
like a tree
889
00:43:02,333 --> 00:43:04,333
and still even if it's not perfect
890
00:43:04,333 --> 00:43:06,166
is going to work believe me
891
00:43:06,166 --> 00:43:08,799
or if you want to increase more contrast
892
00:43:08,966 --> 00:43:09,899
be my guest
893
00:43:09,900 --> 00:43:12,766
and increase the number in here
894
00:43:13,100 --> 00:43:16,266
the nice thing about the role of contrast
895
00:43:16,266 --> 00:43:19,766
is keeping your values between 0 and 1
896
00:43:19,866 --> 00:43:21,733
okay and in here
897
00:43:21,733 --> 00:43:26,533
you can also change the distribution of that center
898
00:43:26,533 --> 00:43:32,466
so you can go to a higher value or reduce
899
00:43:32,466 --> 00:43:35,099
just to make everything really hard
900
00:43:35,100 --> 00:43:38,466
okay so this could be understood
901
00:43:38,766 --> 00:43:42,266
as the center of the solarization
902
00:43:42,266 --> 00:43:46,566
so this is where is going to be your peak okay
903
00:43:48,300 --> 00:43:53,366
so now that we have a hard edge for every grain
904
00:43:53,800 --> 00:43:56,233
we can now copy to the original
905
00:43:56,366 --> 00:43:58,966
and create the primal duplication
906
00:43:58,966 --> 00:44:01,866
so see the difference now
907
00:44:02,333 --> 00:44:05,666
okay so the difference is now mainly
908
00:44:06,500 --> 00:44:11,533
that you are randomizing those soft areas
909
00:44:11,533 --> 00:44:14,499
okay so when you apply that at the end
910
00:44:14,566 --> 00:44:16,299
as you can see here on top
911
00:44:17,200 --> 00:44:18,600
you're not going to see
912
00:44:18,866 --> 00:44:21,299
where the grain is going to mix with the other
913
00:44:21,300 --> 00:44:23,966
because it's actually not blending
914
00:44:23,966 --> 00:44:28,799
it's actually grain on top of other grain okay um
915
00:44:28,966 --> 00:44:32,633
this is the second process of adding grain
916
00:44:32,933 --> 00:44:35,166
to the soft areas
917
00:44:35,400 --> 00:44:37,866
you can go the other way around
918
00:44:38,100 --> 00:44:42,600
and he's making your soft areas super hard
919
00:44:42,733 --> 00:44:46,699
and then apply the grain to all the hard areas
920
00:44:46,700 --> 00:44:49,033
and put everything on top
921
00:44:49,133 --> 00:44:51,799
it's going to be easier if I show you here
922
00:44:52,266 --> 00:44:56,199
so what I did is exactly the same patch okay
923
00:44:56,300 --> 00:45:01,700
and in here I have exactly my same soft patch
924
00:45:02,133 --> 00:45:04,199
what I'm doing is copying
925
00:45:04,200 --> 00:45:05,400
prim multiplying
926
00:45:06,166 --> 00:45:10,466
but what I did now is
927
00:45:10,466 --> 00:45:13,099
okay in this case is going to be a bit silly
928
00:45:13,100 --> 00:45:15,666
because I just can't take this
929
00:45:16,133 --> 00:45:19,399
in my rotor and increase the feather fall off
930
00:45:19,400 --> 00:45:22,700
but I'm doing this even for other cases
931
00:45:22,733 --> 00:45:27,166
when you cannot just change the softening of your patch
932
00:45:27,166 --> 00:45:28,599
so in that case
933
00:45:28,600 --> 00:45:32,066
what I'm doing is mixing as I did before
934
00:45:32,066 --> 00:45:34,199
but instead of mixing with the original
935
00:45:34,200 --> 00:45:37,433
I'm going to mix with the V noise option
936
00:45:37,466 --> 00:45:41,366
okay so now I have my patch without noise
937
00:45:41,366 --> 00:45:44,933
be mixed with my original image without noise
938
00:45:44,933 --> 00:45:47,799
so we are at the same level right now
939
00:45:47,933 --> 00:45:50,566
um remember what I show you before
940
00:45:50,733 --> 00:45:55,433
about the difference of the original
941
00:45:55,500 --> 00:45:57,233
and the original with the effect
942
00:45:57,266 --> 00:46:01,699
well now I'm going to show you another way of isolating
943
00:46:01,766 --> 00:46:03,966
what is called the active area
944
00:46:04,333 --> 00:46:05,999
with this representing
945
00:46:06,000 --> 00:46:11,466
all the pixels change before and after a certain merge
946
00:46:11,566 --> 00:46:15,566
so in here I have this merge operation
947
00:46:15,600 --> 00:46:20,433
which is mixing the image without grain and the patch
948
00:46:20,533 --> 00:46:23,366
so if you want to isolate all the pixels
949
00:46:23,366 --> 00:46:25,233
that I am modifying
950
00:46:26,366 --> 00:46:30,599
from my patch to the V noise option in here
951
00:46:31,333 --> 00:46:34,433
you're going to use this combo of three nodes
952
00:46:34,666 --> 00:46:36,866
which is the difference
953
00:46:37,800 --> 00:46:39,400
which is representing
954
00:46:39,866 --> 00:46:41,166
in every channel
955
00:46:41,200 --> 00:46:42,900
everything that change
956
00:46:43,133 --> 00:46:44,066
as you can see
957
00:46:44,066 --> 00:46:47,466
there are pictures that are change more than others
958
00:46:48,200 --> 00:46:49,500
and now what I'm doing is
959
00:46:49,500 --> 00:46:52,300
I don't want to preserve the difference
960
00:46:52,366 --> 00:46:54,066
for every channel
961
00:46:54,066 --> 00:46:57,266
so for me every pixel that has change
962
00:46:57,266 --> 00:46:59,266
is important as a whole
963
00:46:59,266 --> 00:47:00,599
not for channels
964
00:47:00,766 --> 00:47:02,933
so I'm going to reduce the saturation
965
00:47:02,933 --> 00:47:06,366
but this time I'm going to use another luminance math
966
00:47:06,366 --> 00:47:08,933
which is the maximum okay
967
00:47:08,933 --> 00:47:12,299
the maximum is going to preserve for every channel
968
00:47:12,300 --> 00:47:13,700
the maximum value
969
00:47:13,700 --> 00:47:16,700
okay so now at the end
970
00:47:16,700 --> 00:47:20,033
what I'm going to put is a grade in here
971
00:47:20,766 --> 00:47:21,599
just for the alpha
972
00:47:21,600 --> 00:47:24,266
which is the channel that I'm aiming for
973
00:47:25,066 --> 00:47:26,499
and in there
974
00:47:26,500 --> 00:47:29,000
what I put is the white point
975
00:47:29,066 --> 00:47:31,099
to be at zero
976
00:47:31,333 --> 00:47:34,399
that is going to create that
977
00:47:34,400 --> 00:47:35,633
every pixel
978
00:47:36,100 --> 00:47:38,300
that is not at zero
979
00:47:39,100 --> 00:47:41,066
is going to be white
980
00:47:42,100 --> 00:47:43,466
okay in here
981
00:47:43,466 --> 00:47:44,133
as you can see
982
00:47:44,133 --> 00:47:47,899
there is a certain kind of antialius okay
983
00:47:47,900 --> 00:47:48,933
and that is normal
984
00:47:48,933 --> 00:47:50,366
even in other cases
985
00:47:50,366 --> 00:47:51,633
you will find that there are
986
00:47:51,800 --> 00:47:54,966
areas of some kind of semi transparency
987
00:47:54,966 --> 00:47:56,199
if I reduce
988
00:47:56,200 --> 00:47:59,966
you will see that your image is now changing
989
00:47:59,966 --> 00:48:04,299
so what I want is all my pixels to be
990
00:48:04,566 --> 00:48:06,699
or white or black
991
00:48:06,700 --> 00:48:07,866
nothing in the middle
992
00:48:07,866 --> 00:48:09,866
okay so how
993
00:48:09,866 --> 00:48:11,899
well you can use the gamma
994
00:48:12,000 --> 00:48:15,000
and there is some number you can use in this gamma
995
00:48:15,000 --> 00:48:19,400
which is in here one
996
00:48:20,600 --> 00:48:25,000
with one two 3 4 5 six
997
00:48:25,133 --> 00:48:28,266
7 and eight zeros
998
00:48:28,533 --> 00:48:30,766
why eight zeros
999
00:48:30,966 --> 00:48:33,299
you know it's because of the precision
1000
00:48:33,766 --> 00:48:39,033
of the 32 bit froaking point calculations within milk
1001
00:48:39,400 --> 00:48:42,166
the key Valentina decimal representation
1002
00:48:42,366 --> 00:48:46,366
of a step between
1003
00:48:47,500 --> 00:48:50,633
value of the 32 bit floating point
1004
00:48:50,800 --> 00:48:57,566
is around at 0.7 zeros of precision and one value
1005
00:48:57,600 --> 00:49:02,333
so that if we revert that effect using the gamma
1006
00:49:02,333 --> 00:49:05,899
if we put one with eight zeros
1007
00:49:05,900 --> 00:49:08,633
we are going to increase the gamma enough
1008
00:49:08,666 --> 00:49:11,599
just to cover all the steps in there
1009
00:49:11,600 --> 00:49:12,566
okay of course
1010
00:49:12,566 --> 00:49:14,299
if you use more
1011
00:49:14,300 --> 00:49:16,500
you are going to have the same result
1012
00:49:16,500 --> 00:49:18,333
if you have less probably
1013
00:49:18,333 --> 00:49:21,199
you are not going to have a quite visible result
1014
00:49:21,200 --> 00:49:23,333
but it's going to be the same there
1015
00:49:23,333 --> 00:49:25,266
okay so here you are
1016
00:49:25,266 --> 00:49:26,666
once you apply that
1017
00:49:26,966 --> 00:49:31,999
all your pixels are 0 or 1
1018
00:49:32,100 --> 00:49:33,266
let's look at the alpha
1019
00:49:33,533 --> 00:49:36,266
okay so I copied there
1020
00:49:36,933 --> 00:49:39,099
remember that we are working with a mat
1021
00:49:39,100 --> 00:49:42,200
so black clamp and white clamp
1022
00:49:42,900 --> 00:49:44,266
now we copy there
1023
00:49:44,533 --> 00:49:48,499
we apply the grain through everything okay
1024
00:49:49,100 --> 00:49:54,666
and now we pre multiply using that portion okay
1025
00:49:55,333 --> 00:49:59,166
then we merge everything together
1026
00:49:59,400 --> 00:50:00,500
and here you are
1027
00:50:00,500 --> 00:50:04,266
you are not going to get any step from the grain
1028
00:50:04,766 --> 00:50:05,799
and the patch
1029
00:50:06,333 --> 00:50:08,866
the proling with this method is
1030
00:50:08,866 --> 00:50:11,699
you are going to add areas
1031
00:50:12,800 --> 00:50:15,800
from the regional that has changed
1032
00:50:15,800 --> 00:50:19,000
let me show you with this area of action I created
1033
00:50:19,000 --> 00:50:22,500
this is quite similar to the isolation I did in here
1034
00:50:22,500 --> 00:50:23,600
for the difference
1035
00:50:23,600 --> 00:50:24,700
but in here
1036
00:50:24,733 --> 00:50:27,633
what I did is using a clamp okay
1037
00:50:27,700 --> 00:50:29,766
so if you use a clamp
1038
00:50:29,766 --> 00:50:31,199
and you enable everything
1039
00:50:31,200 --> 00:50:34,200
and you have values that are 0 or 1
1040
00:50:34,200 --> 00:50:36,400
so everything that is not 0
1041
00:50:36,400 --> 00:50:37,600
is going to be 1
1042
00:50:37,600 --> 00:50:41,833
okay so is another method of calculating the same thing
1043
00:50:41,966 --> 00:50:44,599
so you are changing all this area
1044
00:50:45,000 --> 00:50:46,100
and probably
1045
00:50:46,466 --> 00:50:47,599
the area you are
1046
00:50:47,966 --> 00:50:50,766
just aiming to change is the area of the eye
1047
00:50:50,766 --> 00:50:51,599
like in here
1048
00:50:51,600 --> 00:50:55,000
so I'm adding in here an area
1049
00:50:55,733 --> 00:50:58,566
that is being slightly soft
1050
00:51:00,333 --> 00:51:01,133
see
1051
00:51:02,466 --> 00:51:03,899
by the d noise
1052
00:51:04,100 --> 00:51:06,500
so this is the only inconvenient
1053
00:51:06,500 --> 00:51:10,100
I find when I'm using this method
1054
00:51:10,100 --> 00:51:11,533
okay it's very good
1055
00:51:11,533 --> 00:51:12,466
it's gonna work
1056
00:51:12,466 --> 00:51:14,933
it's going to be practically not noticeable
1057
00:51:14,933 --> 00:51:16,066
if well done
1058
00:51:16,733 --> 00:51:19,399
but it's gonna work very well
1059
00:51:19,600 --> 00:51:22,266
when you have a soft image
1060
00:51:22,266 --> 00:51:25,266
or your patch is mixing with a soft area
1061
00:51:25,466 --> 00:51:29,366
but as soon as you get any kind of detail in there
1062
00:51:29,666 --> 00:51:33,799
that detail is going to be slightly lost
1063
00:51:34,066 --> 00:51:36,399
okay so nothing major
1064
00:51:36,400 --> 00:51:38,766
but something to keep in mind
1065
00:51:38,800 --> 00:51:42,533
okay so this area of action we just discuss
1066
00:51:42,533 --> 00:51:46,366
could be a good way of interacting with the image
1067
00:51:46,366 --> 00:51:49,933
just to understand what areas of the image has changed
1068
00:51:49,933 --> 00:51:53,866
so for quality control or quality checks
1069
00:51:54,200 --> 00:51:55,300
you can understand
1070
00:51:55,300 --> 00:51:58,500
what areas of the image you actually change
1071
00:51:58,533 --> 00:52:02,199
if you are creating a rig removal
1072
00:52:02,533 --> 00:52:05,399
and then you find yourself that you are changing areas
1073
00:52:05,400 --> 00:52:07,900
that are not supposed to be changed
1074
00:52:08,000 --> 00:52:10,200
that is something to address
1075
00:52:10,200 --> 00:52:11,566
so in this area
1076
00:52:11,566 --> 00:52:12,966
everything is working fine
1077
00:52:12,966 --> 00:52:14,699
because I was changing the eye
1078
00:52:14,700 --> 00:52:16,766
and is just changing that area
1079
00:52:16,766 --> 00:52:19,666
and also those markers that were in
1080
00:52:19,666 --> 00:52:21,233
in there okay
1081
00:52:21,600 --> 00:52:25,300
so this is the difference from the regional
1082
00:52:26,800 --> 00:52:31,700
okay so this is a good companion of methodologies
1083
00:52:31,700 --> 00:52:34,100
to mixing the grain um
1084
00:52:34,466 --> 00:52:36,699
don't underestimate the grain
1085
00:52:36,700 --> 00:52:39,633
because it's a process that is quite complex
1086
00:52:39,666 --> 00:52:41,799
and even if you pay a lot of attention
1087
00:52:41,800 --> 00:52:44,033
to many aspects of the CG
1088
00:52:44,133 --> 00:52:48,733
um the integration and lighting and everything
1089
00:52:48,733 --> 00:52:51,499
but then your grain fails
1090
00:52:51,533 --> 00:52:53,466
everything is gonna fail
1091
00:52:53,600 --> 00:52:58,600
okay so I hope you find this class interesting
1092
00:52:58,666 --> 00:53:00,999
um that's all for this week
1093
00:53:01,066 --> 00:53:02,299
if you have any comments
1094
00:53:02,300 --> 00:53:04,300
questions or suggestions
1095
00:53:04,300 --> 00:53:06,300
please post them in the forums
1096
00:53:06,566 --> 00:53:08,599
and until next week
1097
00:53:08,600 --> 00:53:12,500
this has been Victor Perez for FX PhD
1098
00:53:12,866 --> 00:53:14,033
thanks for watching
78182
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.